* cdtest-foo.cc: Use explicit "#pragma implementation".
* cdtest-bar.cc: Renamed from cdtest-func.cc.
* Makefile.in: References to cdtest-func.o changed to cdtest-bar.o.
* ldlang.c (out_bfd_get_section_by_name): Remove.
(wild_section): Call bfd_get_section_by_name rather than
our_bfd_get_section_by_name. Don't call wild_doit if there is no
section.
(lang_create_output_section_statements): Remove.
(map_input_to_output_sections): For several cases, call init_os if
it has not already been called.
(lang_size_sections): If output section was not created, skip it.
(lang_process): Don't call lan_create_output_section_statements.
(lang_place_orphans): Skip files with just_syms_flags set to true.
* ld.texinfo: Document change.
* emultempl/gld960.em, emultempl/gld960c.em, emultempl/lnk960.em,
emultempl/hppaelf.em, emultempl/m88kbcs.em, emultempl/vanilla.em: Ditto.
* Makefile.in: Changed all generated file names.
(ldemul-list.h): Depend on Makefile, not config.status. Changed sed patterns
to handle new filenames.
* ldlang.c (lang_size_sections): If dot moves because of an
assignment, don't try to insert a pad into the absolute output
section, just change the address of the default memory region
instead.
* Makefile.in (em_m68klynx.c, em_i386lynx.c, em_sparclynx.c): Use
Lynx-specific script templates.
* configure.in (sparclite*-*-coff): Use coff-sparc.
* emulparams/i386lynx.sh (SCRIPT_NAME): Set to i386lynx.
* emulparams/sparclynx.sh (SCRIPT_NAME): Set to sparclynx.
(ENTRY): Set to __main.
* scripttempl/i386lynx.sc: New file, script for I386 Lynx.
* scripttempl/m68klynx.sc: Add insertion of ctor/dtor sections.
* scripttempl/sparclynx.sc: New file, script for uSparc Lynx.
.fini sections into .text; keep them separate.
* config/i386sco.mh (HOSTING_CRT0): If ../gcc/crtbegin.o does not
exist, get crtbegin based on gcc -print-libgcc-file-name.
(HOSTING_LIBS): Similar change for ../gcc/crtend.o.
real to true for lang_input_file_is_marker_enum. Clear the_bfd.
(lang_add_input_file): Pass true to new_afile for add_to_list.
(lookup_name): Remove force_load argument. Changed all callers.
Pass false to new_afile for add_to_list. Split loading of symbols
out into separate function.
(load_symbols): New function split out of lookup_name. Don't load
the symbols if they are already loaded.
(open_input_bfds): For lang_input_statement_enum call load_symbols
rather than lookup_name.
(lang_process): Pass abs_output_section rather than NULL to
lang_size_sections.
(lang_startup): Set real field of first_file to true.
* ldmain.c (add_to_set): Now takes reloc argument rather than
bitsize. Check config.build_constructors here. If an new hash
table entry is created, mark it as undefined.
(constructor_callback): No longer takes bitsize argument. Pass
BFD_RELOC_CTOR to ldctor_add_set_entry, but first make sure the
BFD backend supports it.
(reloc_overflow): Handle a NULL abfd argument.
(reloc_dangerous, unattached_reloc): Likewise.
* ldctor.c: Include ldmain.h.
(struct set_info): Change bitsize field to reloc.
(ldctor_add_set_entry): Now takes reloc argument rather than
bitsize. Don't bother to check config.build_constructors here.
(ldctor_build_sets): Get the size from the reloc howto. If
generating relocateable output, call lang_add_reloc rather than
lang_add_data.
* ldctor.h (ldctor_add_set_entry): Change declaration to use reloc
instead of bitsize.
* ldlang.h (statement_enum): Add lang_reloc_statement_enum.
(lang_reloc_statement_type): New structure.
(lang_statement_union_type): Add reloc_statement field.
(lang_add_reloc): Declare new function.
* ldlang.c (lang_for_each_statement_worker): Handle
lang_reloc_statement_enum.
(map_input_to_output_sections, print_statement): Likewise.
(lang_size_sections, lang_do_assignments): Likewise.
(print_reloc_statement): New function.
(lang_add_reloc): New function.
* ldwrite.c (build_link_order): Handle lang_reloc_statement_enum.
Ian Lance Taylor